A nonorthogonal passive decoy-state method is presented with a reconstructive weak coherent state source. The method dose not prepare decoy states actively and divides the receiver detection events into two groups, i.e., triggered components and nontriggered components, according to triggering situation of the sender detector. Both triggered and nontriggered components, as signal states and decoy states, are used to do some estimations and to generate secure key. The simulation results show that a better key generation rate and a longer secure transmission distance can be obtained with the nonorthogonal passive decoy-state method than with the existing passive methods, and that the performance is comparable to the theoretical limit of an active infinite decoy-state protocol. Furthermore, the nontriggered component contribution to key generation offsets the limitation of the detector low efficiency, and the performance of the method dose not depend on the detector efficiency of sender. Because decoy states need not be prepared actively, and our protocol is easy to implement and apply to quantum key distribution at high transmission rates.